--- hide: - toc --- # Create a MinIO instance 1. Select __Minio Storage__ from the left navigation bar. 2. You can click the __New Instance__ button in the upper right corner of the list. > If you are using it for the first time, you need to select a workspace first and click __Deploy Now__ to create a MinIO instance. 3. Fill in the basic information of the instance by referring to the information below, and then click __Next__ . !!! note - The instance name and cluster/namespace cannot be modified after the instance is created - Pay attention to check the filling requirements below the input box, and enter the content that meets the requirements - If the installation environment test fails, you can follow the prompts to install the relevant plug-ins and re-create the instance, or check __Do not need to monitor__ and proceed directly to follow-up operations 4. Fill in the configuration specifications with reference to the information below, and then click __Next__ . - The deployment mode cannot be changed after the instance is created - High availability deployment mode is recommended in production mode - At least 4 replicas are required in high availability mode - Storage class: The selected storage class should have enough available resources, otherwise the instance creation will fail due to insufficient resources - Storage capacity: how much capacity each disk has. **Cannot be lowered after instance creation** - Disks per copy: How many secondary disks to provide for each copy. **Cannot be lowered after instance creation** 5. Fill in the service settings with reference to the information below, and click __Next__ . - Intra-cluster access: services can only be accessed within the same cluster - Node port: access services through node IP and static ports, and support access to services from outside the cluster - Load balancer: use the cloud service provider's load balancer to make the service publicly accessible - Load Balancer/External Traffic Policy: Specifies whether the service routes external traffic to node-local or cluster-wide breakpoints - Cluster: Traffic can be forwarded to Pods on other nodes in the cluster - Local: Traffic can only be forwarded to Pods on this node - Console account: the username and password needed to access this newly created instance ??? note "Click to view advanced configuration instructions" - Bucket name: create a new bucket under this instance, and set the name of the new bucket - Scheduling policy/condition: set the node affinity of Pod scheduling, refer to the official Kubernetes document [node affinity](https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod -node/#affinity-and-anti-affinity) - Satisfy as much as possible: try to schedule to the nodes that meet the rules first. Pod scheduling is also performed if no matching node is found - Must be satisfied: Pod scheduling is only performed when a node satisfying the rules is found - Scheduling strategy/weight: Set weights for nodes that meet each scheduling strategy, preferably using a strategy with a high weight. Value range 1 to 100 - Scheduling strategy/selector - In: The node must contain the selected label, and the value of the label must **belong** to a set of values. Multiple values ​​are separated by __;__ - NotIn: The node must contain the selected label, and the value of the label must **not belong to a value set. Multiple values ​​are separated by __;__ - Exists: It is enough for the node to contain a certain label, and do not pay attention to the specific value of the label - DoesNotExists: The node does not contain a certain label, and does not pay attention to the specific value of the label - Gt: The node must contain a certain label, and the value of the label must be greater than a certain integer - Lt: The node must contain a certain label, and the value of the label must be less than a certain integer 6. Confirm that the instance configuration information is correct, and click __OK__ to complete the creation. 7. Return to the instance list page to check whether the instance is successfully created. > The status of the newly created instance is __Not Ready__ , and the status becomes __Running__ after all related containers are successfully started.